Conservation recommendations for Nee Soon 
209 
downstream, potentially reducing water depths in larger streams and/or changing 
benthic habitats towards more, finer, “softer” sediments further downstream. In concert 
with research to investigate this, trails of “soft-engineering” approaches conducted 
initially off-site could be used to identify i) suitable plants; and ii) suitable techniques 
to mitigate against stream bank erosion. 
Cryogenic collection, imaging and barcoding 
Being able to identify specimens to species level is important for most in-depth study 
of biological systems. However, getting these identifications is particularly challenging 
in tropical environments. Fortunately, a number of new tools promise to make this 
task less daunting (Kutty et al., 2018). New imaging techniques help with illustrating 
relevant characters and new and cheaper DNA barcoding techniques will allow for the 
generation of databases that can be used by many researchers. 
Making the fauna and flora of Nee Soon freshwater swamp forest and of 
Singapore identifiable is achievable. The samples that have been collected and stored 
have the potential to reveal the presence of several hundred or even thousands of 
species. By focusing on particular taxa belonging to different ecological guilds, it is 
feasible to begin understanding species turnover rates across habitats in Singapore and 
to use this information for conserving Singapore’s native fauna and flora. A particularly 
high priority should be using the newly developed plant barcoding techniques for all 
of Singapore’s vascular plant species. This will allow for in-depth studies of species 
interactions between plants and animals (e.g. pollination). 
Eco-hydrological modelling 
An integrated eco-hydrological model was developed in this study for the Nee 
Soon freshwater swamp forest using Milce-SHE (Sun et al., 2018). The Mike-SHE 
model simulates various water flow processes in the hydrological cycle, such as 
rainfall, reservoir water of 3 reservoirs (Upper Seletar, Upper Peirce, Lower Peirce), 
evapotranspiration, overland flow, infiltration, and groundwater flow. 
The surveyed GIS data, including the stream network, the cross-sections and 
the updated DEM, were incorporated in the model setup to make the model more 
representative. The spatial and temporal variations of leaf area index (LAI) and 
reference evapotranspiration (ET) retrieved from the remote sensing data, with the 
aid of the root depth (RD) information from Vegetation Ecology team, were used 
to establish a two-layer water balance model to account for the water loss from 
evapotranspiration and the amount of water recharging to the saturated zone. In 
addition, the field measurements from piezometers and stream sondes were processed 
and integrated to calibrate and validate the model parameters. 
A swamp forest extent map was derived from the numerical model simulation, 
following the definiton of groundwater table shallower than 0.2 m below the surface 
level. The model’s simulated swamp forest extent matches rather well with the swamp 
forest map resulting from the study of O’Dempsey & Chew (2013), with an increased 
area of 140 ha from 111 ha. The Nee Soon freshwater swamp forest was divided into 8 
sub-catchments based on catchment delineation according to the topographic features. 
